[ Question ]

How Much Does New Research Inform Us About Existential Climate Risk?

by zdgroff1 min read22nd Jul 20205 comments


A new study came out today that seems to be a pretty big deal, though it's not my area of expertise. A team of scientists narrowed estimates of climate sensitivity. In The Precipice, Toby Ord notes that uncertainty about things like this increases the probability we should place on such risks. Does anyone know how much this study should lower our credence (if at all)?

Here are the "Key Points" from the study:

● We assess evidence relevant to Earth’s climate sensitivity S: feedback process understanding, and the historical and paleo-climate records.
● All three lines of evidence are difficult to reconcile with S < 2 K, while paleo evidence provides the strongest case against S > 4.5 K. 48
● A Bayesian calculation finds a 66% range of 2.6-3.9 K, which remains within the bounds 2.3-4.5 K under plausible robustness tests.
New Answer
Ask Related Question
New Comment

1 Answers

Thanks for posting this. This does seem to correct a lot of the common stated problems with estimates of S by incorporating all the lines of evidence. It'll be interesting to see how this is received in AR6.

I have updated the guesstimate model from my How Hot Will it Get piece to reflect the findings here. The scenarios are labelled as the WCRP estimates of climate sensitivity.

Overall, I don't view this as especially good news.

  • The 95% confidence interval for S for the baseline case is 2.3K to 4.7K. Depending on the emissions scenario you put in, this implies a 5% chance of >6.3K-7K of warming on pre-industrial.
  • With robustness checks, the 95% CI for S is 2 to 5.7K. Depending on the emissions scenario, this implies a 5% chance of >7K-10K on pre-industrial.

From what I have read of their assumptions, the Baseline case seems more plausible, e.g. the robustness checks include an assumption of a uniform prior over S, which seems wrong. This suggests that the high chance of extreme warming suggested by Wagner and Weitzman is less likely. Still, the chance of >6K is way too high.